Final Activity Report Summary - MULTNEURIMAGNETWORK (Multimodal fMRI/EEG/MEG study of large-scale functional integration) Until recently, the mapping of human brain function consisted of identifying brain regions that were associated with the treatment of a cognitive task. Over the past two decades, this was the (very productive) primary preoccupation of the neuroimaging research community. Pioneering the discipline, the group of Pr. K. J. Friston made a significant effort to extend the standard paradigm to a more mechanistic view of brain processes. In this view, brain function emerges from the flow of information among brain areas. Accordingly, novel mathematical models of and associated statistical techniques are required to characterising the quantity and the nature of this information, given neuroimaging measurements of brain activity.During the two years of the Marie-Curie fellowship, significant advances in modelling have been achieved. Specifically, we have proposed a number of neurobiologically realistic dynamic causal models of neuroimaging data. Based on these models, we have developed probabilistic data analysis methods that aim at identifying the structure of brain networks engaged in the treatment of any specific cognitive task. We have now started to apply these techniques to a series of neuroscientific investigations of human perception, learning and decision making. Preliminary results indicate a tight relationship between quick modifications of the large-scale network connectivity (short-term cerebral plasticity) and learning. A deeper understanding of this relationship has disclosed new modelling avenues, potentially bridging the gap between neurobiological brain processes and basic cognitive functions.
